Application of multi temporal satellite images for improvement of maize phenology models prediction

Crop phonological stages are commonly predicted by using accumulated growth degree days(AGDD).In this study a combined model of AGDD and remotely sensed NDVI has been developed for prediction of maize (cv. K407) phenology in Karaj using a nine year (2002 to 2010) dataset. For smoothing the existing noises of image processing, a combination of double logistic and weighing average (DL-WLS) approaches was employed. The results of combined phenology model were compared by two frequently used methods based on AGDD and date of sowing. The findings showed that in general, the developed model predicted the first 7 phenological stages of emergence to milky, more accurately comparing to other approaches (with average 2 and 2.5 days difference with observed dates, respectively) but was inaccurate for maturity stage. Our study highlights the need for further improvements in observations in the region.